An Intrusion Detection Model Using Machine Learning for Safeguarding Internet of Things Infrastructure Against Cyber Threats

The Internet of Things (IoT) has become a critical enabler of modern digital services, yet its rapid growth has exposed billions of devices to cyber threats such as denial-of-service (DoS), distributed denial-of-service (DDoS), malware, and man-in-the-middle attacks. This study develops a machine learning-based Intrusion Detection System (IDS) tailored for IoT infrastructure security. Three benchmark datasets—BoT-IoT, IoT Healthcare, and TON-IoT—were preprocessed through exploratory data analysis, feature selection using ANOVA and Logistic Regression, and dimensionality reduction via PCA. Four models were implemented and optimized: Random Forest (RF), XGBoost, Recurrent Neural Network (RNN), and Gaussian Naive Bayes (GNB). Evaluation metrics included accuracy, precision, recall, and F1-score, with datasets split 70-15-15 for training, validation, and testing. Results indicate that RF consistently achieved the best accuracy (93–94%) across datasets, while XGBoost delivered comparable performance with shorter training time. RNN showed moderate performance, and GNB lagged due to its simplifying assumptions. The findings highlight that robust, scalable IDS solutions can be developed for IoT ecosystems, ensuring confidentiality, integrity, and availability